Amino acid mediated borane reduction of ketones II

Article abstract of DOI:10.1007/s007060270003

Acetophenone, 2,2-dimethylcyclopentanone, 3,3-dimethyl-2-butanone, 3-methyl-2-butanone, and 2-pentanone were reduced with borane mediated by (S)-alanine, (S)-methionine, (S)-leucine, (S)-valine, and (S)-isoleucine in very good yields giving predominantly alcohols of (R)-configuration (ee = 23-89%). A molecular topology based model was developed for describing the influence of the substituents, both in the oxazaborolidine type reagent and in the ketone, on the observed chiral induction.

Full text of DOI:10.1007/s007060270003

Monatshefte fuÈr Chemie 133, 23±29 (2002)
Amino Acid Mediated
Borane Reduction of Ketones IIa
Ã
ꢀ
Aleksandar V. Teodorovic , Milan D. Joksovic, Ivan Gutman,
ꢀ
ꢁ
and Zeljko Tomovic
ꢀ
Faculty of Science, University of Kragujevac, YU-34000, Kragujevac, Yugoslavia
Summary. Acetophenone, 2,2-dimethylcyclopentanone, 3,3-dimethyl-2-butanone, 3-methyl-2-
butanone, and 2-pentanone were reduced with borane mediated by (S)-alanine, (S)-methionine,
(S)-leucine, (S)-valine, and (S)-isoleucine in very good yields giving predominantly alcohols of (R)-
con®guration (ee ˆ 23±89%). A molecular topology based model was developed for describing the
in¯uence of the substituents, both in the oxazaborolidine type reagent and in the ketone, on the
observed chiral induction.
Keywords. (S)-Amino acids; Ketones; Alcohols; Chiral induction; Molecular topology based model.
Introduction
Among the catalytic enantioselective borane reductions, the oxazaborolidine medi-
ated reaction, introduced by Corey [1] who followed the pioneering work by Itsuno
[2], stands as a well-established methodology for the synthesis of chiral second-
ary alcohols [3]. The active species for hydride transfer is an oxazaborolidine-borane
adduct [1] which can be employed either stoichiometrically or catalytically [4].
Borane reductions of prochiral ketones with catalytic amounts of (S)-proline in
re¯uxing toluene gave the corresponding alcohols in good enantiomeric excess
(ee ˆ 81±95%) [5]:
R1COR2 ‡ ꢀS†-amino acid ꢀRÀCHꢀNH2†COOH†
‡ BH3ÁTHF ! R1CHꢀOH†R2
ꢀ1†
In continuation of the investigation of the reaction of Eq. (1) we have performed
borane reductions of the same or similar ketones using (S)-proline, (S)-phenyl-
alanine, and (S)-valine in THF at room temperature [6]. The results obtained
[6] were of limited preparative value, i.e. had small ee-values. We now report
on further borane reductions of ketones involving (S)-alanine, (S)-methionine, (S)-
leucine, (S)-valine, and (S)-isoleucine which, at least in some cases, exhibit a
higher degree of chiral induction.
a
For part I, see Ref. [6a]
Ã
Corresponding author. E-mail: alek@knez.uis.kg.ac.yu

ꢀ
A. V. Teodorovic et al.
24
In addition, we put forward a simple topological model capable of predicting
the in¯uence of the structure of both the reducing agent (i.e. of the group R in the
amino acid in Eq. (1)) and the ketone (i.e. of the groups R1 and R2 in Eq. (1)) on the
observed chiral inductions.
Results and Discussion
When we used proline as a chiral modi®er in the reduction of acetophenone [6a],
the best molar proline:borane:acetophenone ratio was estimated to be 1:7:10 (85%
ee of 1-phenylethan-1-ol; other conditions: (i) BH3ÁTHF or BH3ÁSMe2 as the
source of borane; (ii) THF as solvent; (iii) 10 min of preparation of the reagent at
room temperature; (iv) 10 min of reduction of ketone at room temperature). This
ratio is in agreement with two earlier results obtained by Corey et al. [1a] in the
reduction performed in presence of diphenyl-3-oxa-1-aza-2-borabicyclo [3.3.0]-
octane. Surprisingly, in the variant with slower addition of ketone (80% ee;
reaction time: 2.5 h) [6b], the optimal valine:borane:acetophenone ratio was found
to be 1:2:1. The reaction was still catalytic, because as much as 89% of (S)-valine
could be recovered; consequently, only 4% of valinole was isolated. The other
amino acids applied in this work have showed a similar behaviour towards the
THF/borane system. The reaction mixture prevails in the form of suspension. It
should be stressed that the reagent prepared directly from valinol [7] is completely
soluble in THF. In order to obtain mutually comparable results, the same reaction
conditions were applied for all amino acids used in this work. The obtained results
are given in Table 1.
The role of the amino acids in the borane reduction of prochiral ketones
(R1COR2, R1 ˆ R2) might be accounted for by formation of the corresponding
oxazaborolidine (1; Scheme 1). Through the adjacent boron and nitrogen atoms of
1, one ketone and one borane molecule can react by the double docking mechanism
Table 1. Enantiomeric excess (ee/%) of the alcohols obtained by reduction of ketones with borane
modi®ed with (S)-amino acids; molar amino acid:BH3 Á SMe2:ketone ratio ˆ 1:2:1; reagent prepara-
tion time: 2.5 h; conversion of the ketones: ꢁ100%; chemical purities of the alcohols after distillation:
>99%; in all cases, the (R)-con®guration of the alcohols was predominant, as inferred by the sign of
rotation; the reagent prepared from phenylalanine gave, under same conditions, 1-phenylethan-1-ol of
73% ee; a 1:7:10 phenylalanine:borane:acetophenone mole ratio gave, in 10 min of reaction time, 1-
phenylethan-1-ol in 74% ee [6a]; % of amino acid recovered: 87 (Ala), 82 (Met), 88 (Leu), 89 (Val ), 87
(Ile); % of aminols recovered: 2 (Ala), 7 (Met), 3 (Leu), 4 (Val ), 3 (Ile); the yield of isolated alcohol
(%) is given in parentheses
(S)-Amino acid employed
Ketone
Alanine
Methionine Leucine
Valine
Isoleucine
2-Pentanone
23 (86)
31 (80)
36 (81)
34 (87)
42 (86)
55 (87)
57 (90)
64 (92)
33 (81)
50 (80)
57 (82)
56 (91)
71 (92)
36 (86)
55 (82)
64 (83)
59 (81)
80 (91)
45 (82)
59 (85)
68 (82)
73 (87)
89 (95)
3-Methyl-2-butanone
3,3-Dimethyl-2-butanone
2,2-Dimethylcyclopentanone 36 (87)
Acetophenone 57 (90)

Asymmetric Reduction of Ketones
25
Scheme 1
[1a]. The favorable six-membered cyclic arrangements (4 and 5; Scheme 1) depend
on the oxygen lone-pair involved in the complexation and the enantiotopic face of
the carbonyl compound exposed to the NBHÀ3 unit. The primary requirement for
obtaining a good oxazaborolidine catalyst capable of achieving high enantiomeric
excess in prochiral ketone reductions is to completely block one face of the
oxazaborolidine [8, 9]. In our study, we varied the alkyl group in structure 1 from a
small one as methyl to a moderately bulky group as sec-butyl. The results in Table
1 show that the bulkiness of the alkyl group in the reagent signi®cantly affects the
enantiomeric excesses of the reaction product. A substituent in the supposed
oxazaborolidine±borane adduct type reagent (which originates from the corre-
sponding amino acid) affects the ratio of the adducts 2 and 3, thus in¯uencing the
formation of 4 and 5 by ketone complexation (Scheme 1).
Modeling the structure dependence of the enantiomeric excess
One serious limitation of molecular graph based structure models of organic
compounds and reactions is that they do not distinguish between stereoisomers.
Recently, efforts have been undertaken to overcome this shortcoming (see Ref. [10]
and references cited therein). In order to describe the in¯uence of the bulkiness of
the alkyl group R in the supposed active form (as above described) of the reducing

ꢀ
A. V. Teodorovic et al.
26
agent as well as of the groups R1 and R2 of the ketone (cf. Eq. (1)) on the resulting
enantiomeric excess, we developed a topological model upon which the following
requirements were imposed:
(a) If R1 ˆ R2 then the predicted ee-value must be zero, irrespective of the nature
of the group R (i.e. irrespective of the amino acid employed)
(b) The predicted ee-value must not change if R1 and R2 exchange place
(c) The effect of the groups R1 and R2 is modeled by the same topological index
denoted by T
Ã
(d) The effect of the group R is modeled by a topological index denoted by T
whose structure dependence is identical or as similar as possible to that of T
Ã
(e) The topological indices T and T depend on the rooted molecular graphs of the
groups R, R1, and R2, with the root corresponding to the position at which R,
R1, and R2 are connected to the remainder of the molecule (cf. Eq. (1))
Conditions (a) and (b) re¯ect obvious stereochemical features, whereas (c) and
(e) are just mathematical reformulations of elementary chemical facts. Condition
(d) is based on the intention to make our approach as simple as possible. Bearing in
mind (a)±(e) we constructed and examined several mathematical models, of which
that of Eq. (2) proved to be the most ef®cient.
Ã
ee ˆ jTꢀR1† À TꢀR2†jꢀaÁT ꢀR† ‡ b†
ꢀ2†
In Eq. (2), a and b are least-squares ®tting parameters, and the topological indices
P
P
Àx
T and T are de®ned as TꢀG†ˆ v dꢀv; rjG† and T ꢀG† ˆ v dꢀv; rjG†Ày.
Both summations go over all vertices of the rooted graph G, whose root is r.
The topological distance between the vertex v and the root r ( ˆ number of edges
in the shortest path connecting v and r) is denoted by dꢀv; rjG†; for details, see Ref.
[11]. The parameters x and y measure the speed by which the in¯uence of structural
details on the examined property (in this case, on ee) diminishes with increasing
distance from the reaction center. These parameters were adjusted so that Eq. (2)
gave optimal agreement with experimentally determined ee-values.
Ã
Ã
We examined the model of Eq. (2) using hydrogen-®lled and hydrogen-
depleted moleculars [10a]; the latter gave slightly better results.
In the case of 2,2-dimethylcyclopentanone it is not self-evident what R1 and R2
should be. Following Raychaudury [12], we associated the carbon atoms closer to
one side of the keto group with R1 and those closer to its other side with R2.
By using all 25 experimental ee-values we obtained a ˆ 252.1, b ˆ À209.5, x ˆ
1.5, and y ˆ 4.0 for the parameters of Eq. (2). With these parameters, the correlation
coef®cient was 0.898, and ee was reproduced with an average error of 7.2%.
Analyzing the results obtained (see Fig. 1) we found that one experimental
result (that for the reduction of 2,2-dimethylcyclopentanone with (S )-isoleucine) is
an outlier. Without this single point (indicated in Fig. 1 by an arrow), the correlation
coef®cient increases to 0.926, whereas the average error remains essentially the
same. By checking the respective experiment it was found that the ee-value reported
above is correct.
As already mentioned, the treatment of 2,2-dimethylcyclopentanone (and, in
general, of any cycloalkanone) by means of our model is somewhat ambiguous.
When avoiding the ®ve experimental data for 2,2-dimethylcyclopentanone we

Asymmetric Reduction of Ketones
27
Fig. 1. Experimental vs. calculated ee-values; for details, see text
obtained a ˆ 263.1, b ˆ À230.9, x ˆ 1.2, y ˆ 4.5, and the model became signi®cantly
more accurate (average error 5.1%, correlation coef®cient 0.962).
Anyway, the model of Eq. (2) is capable of reproducing the ee-values within
less than 10%, which for usual purposes (in particular for the prediction of which
reactions are valuable from a preparative point of view) may be considered as
satisfactory. On the other hand, the model is quite simple and requires a minute
amount of numerical work.
It is interesting to note that we found that the exponent y is signi®cantly larger
(about three times) than the exponent x. This implies that, at least within our
model, only those structural features of the amino acid which are located near the
amino group (i.e. local structural details) have a pronounced effect on the chiral
induction. In other words, the bulkiness of the group R of the amino acid is much
less decisive for the observed degree of chiral induction (measured as ee) than the
bulkiness of the groups R1 and R2 attached to the keto group.
Experimental
All operations concerning air sensitive materials were carried out under Ar [13]. THF was dried over
Ê
4 A molecular sieves and distilled from sodium benzophenone ketyl prior to use. Acetophenone, 2,2-
dimethylcyclopentanone, 3,3-dimethyl-2-butanone, 3-methyl-2-butanone, 10 M borane-methyl sul®de
complex, and (R)-( ‡ )-ꢀ-methoxy-ꢀ-(tri¯uoromethyl)-phenylacetic acid (MTPA) were obtained from
Aldrich. Amino acids were obtained from Fluka and used without puri®cation. Speci®c rotations were
determined on a Perkin-Elmer 241 polarimeter. Analytical gas chromatography (GC) was carried out
on a Hewlett-Packard 5890A gas chromatograph using SPB-5 capillary columns. The alcohols were

ꢀ
A. V. Teodorovic et al.
28
puri®ed by preparative GC on a Carbowax 20 M column. The optical purity of the alcohols was
determined by analysis of the corresponding MTPA esters. The chiral regents were prepared as given
below and used in situ for reduction.
Typical experimental procedure
An oven-dried 100 cm3 round-bottomed ¯ask equipped with a magnetic stirring bar and a septum
was cooled to room temperature in a stream of Ar. (S)-Isoleucine (0.918 g, 7 mmol) was transferred
to the ¯ask in a glove bag, and 15 cm3 of THF were added. Then, 1.4 cm3 of 10 M borane-methyl
sul®de complex was added dropwise via a syringe at a rate of 1 cm3/min. After stirring for additional
2.5 h, 0.84 g acetophenone (7 mmol) in 4.1 cm3 THF was added in portions of 0.5 cm3 every 5 min,
and the resulting reaction mixture was stirred for 1 h at room temperature. The solvent was
evaporated under reduced pressure, and the residue was treated with 10 cm3 of aqueous 3 M NaOH.
When the evolution of H2 was complete, the solution was extracted with ether (2 Â 10 cm3), and the
combined organic phase was washed with 2 M HCl, saturated aqueous NaHCO3, and H2O. The
etheral solution was dried over Na2SO4 and distilled. 1-Phenylethan-1-ol, b.p.: 98ꢂC (20 mm Hg),
was obtained (0.81 g, 95% yield, 99% GC purity). The optical rotation (neat), after further
puri®cation by preparative GC, was determined to be [ꢀ]D ˆ ‡ 38.13 Ref. [14]: [ꢀ]D ˆ ‡ 42.85
(neat) and revealed an excess of the (R)-enantiomer of 1-phenylethan-1-ol. The obtained alcohol was
converted to its MTPA ester using a literature procedure [15] and analyzed by GC to reveal an ee of
89% in the (R)-isomer. The other ketones were reduced in an analogous manner using (S)-isoleucine,
(S)-alanine, (S)-methionine, (S)-leucine, or (S)-valine (Table 1).
(S)-Isoleucine (0.809 g, 88%) was recovered by adjusting the basic aqueous solution (before
extraction with ether) to pH 5.74, ®ltering off, washing with ethanol and ether, and drying. The
®ltrate was combined with the acidic aqueous solution (obtained after washing of the etheral phase
with 2 M HCl) and treated with 2 M NaOH up to pH ꢁ 11, followed by continuous extraction with
ether. After evaporation of the solvent, 3% of (S)-2-amino-3-methylpentan-1-ol (isoleucinol) were
obtained. The other amino acids were treated in a similar manner (Table 1).
Acknowledgements
This research was supported by grants from the Ministry of Sciences and Technology of Serbia.
ꢁ
Special thanks are due to Mr. Velimir Mitroviꢀc, Director of LATEX-Cacak.
ꢁ
References
[1] a) Corey EJ, Bakshi RK, Shibata S (1987) J Am Chem Soc 109: 5551; b) Corey EJ, Bakshi RK,
Shibata S, Chen C-P, Singh VK (1987) J Am Chem Soc 109: 7925
[2] Itsuno S, Ito K, Hirao A, Nakahama S (1983) J Chem Soc Chem Commun 469
[3] a) Deloux L, Srebnik M (1993) Chem Rev 93: 763; b) Singh VK (1992) Synthesis, 605; c) Corey
EJ, Helal CJ (1988) Angew Chem Int Ed Engl 37: 1986
[4] Mathre DJ, Thompson AS, Douglas AW, Hoogsteen K, Carroll JD, Corley EG, Grabowski EJJ
(1993) J Org Chem 58: 2880
[5] Brunel JM, Maffei M, Buono G (1993) Tetrahedron Asymm 4: 2255
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
[6] a) Teodorovic AV, Joksovic MD, Konstantinovic SK, Mojsilovic BM, Mihailovic ML (2000)
Monatsh Chem 131: 91; b) Teodorovic AV, Joksovic MD, Konstantinovic SK, Milovanovic JN,
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
Mojsilovic BM (2000) J Serb Chem Soc 65: 549
[7] Itsuno S, Hirao A, Nakahama S, Yamazaky N (1983) J Chem Perkin Trans I, 1673
[8] Quallich GJ, Woodall TM (1993) Tetrahedron Lett 34: 4145
ꢂ
[9] Puigjaner C, Vidal-Ferran A, Moyano A, Pericas MA, Riera A (1999) J Org Chem 64: 7902

Asymmetric Reduction of Ketones
29
[10] a) Todeschini R, Consoni V (2000) Handbook of Molecular Descriptors. Wiley-VCH, Weinheim;
b) Golbraikh A, BonchevD, Tropsha A (2001) J Chem Inf Comput Sci 41: 147; c) Aires-de-
Sousa J, Gasteiger J (2001) J Chem Inf Comput Sci 41: 369
[11] Gutman I, Polansky OE (1986) Mathematical Concepts in Organic Chemistry. Springer, Berlin
[12] Raychaudury C, Banerjee A, Bag P, Roy S (1999) J Chem Inf Comput Sci 39: 248
[13] Brown HC, Kramer GW, Levy AB, Midland MM (1975) Organic Syntheses via Boranes, chapter
9. Wiley, New York, p 191
[14] Pickard RH, Kenyon J (1911) J Chem Soc 99: 45
[15] Dale JA, Dull DL, Mosher HS (1969) J Org Chem 34: 2543
Received May 31, 2001. Accepted June 25, 2001